Automated Detection of Infant Holding Using Wearable Sensing

Abstract

Physical contact is critical for children’s physical and emotional growth and well-being. Previous studies of physical contact are limited to relatively short periods of direct observation and self-report methods. These methods limit researchers’ understanding of the natural variation in physical contact across families, and its specific impacts on child development. In this study we develop a mobile sensing platform that can provide objective, unobtrusive, and continuous measurements of physical contact in naturalistic home interactions. Using commercially available motion detectors, our model reaches an accuracy of 0.870 (std: 0.059) for a second-by-second binary classification of holding. In addition, we detail five assessment scenarios applicable to the development of activity recognition models for social science research, where required accuracy may vary as a function of the intended use. Finally, we propose a grand vision for leveraging mobile sensors to access high-density markers of multiple determinants of early parent-child interactions, with implications for basic science and intervention.

Longitudinal relations between independent walking, body position, and object experiences in home life.

“…In the future, characterizing infant motor behavior from wearable sensors may be a promising alternative. Recent examples include automatically classifying whether infants are held (Yao et al, 2019), locomoting (Airaksinen et al, 2020), and in different body positions (Franchak et al, 2021; Franchak, Tang, et al, 2023). However, other types of behaviors—such as whether infants are grasping objects—and other aspects of interactions—what kind of object during what type of activity or social situation—may be impossible to decipher from motion alone.…”

Oxytocin but not naturally occurring variation in caregiver touch associates with infant social orienting

“…Even though in our study we observed significant variability in parental touch (as indexed for instance by the distribution of the PICTS scores), it is probable that the measures we used were not sensitive to the extreme ends of the caregiver behaviour spectrum. Emerging technologies, such as devices recording body contact (Yao et al., 2019), could partially address this issue by allowing us to capture touching behaviors over extended periods of time and in infants’ natural environment, and thus might be the future of touch research in infancy. Moreover, one of our main measures of interest, salivary oxytocin, has been associated with some controversies about its validity and specificity (Uvnäs‐Moberg et al., 2020), and it has also yielded a substantial amount of missing data in our study.…”

Automatic assessment of infant carrying and holding using at-home wearable recordings

“…Parental surveys have indicated privacy concerns with in-home audio and video recordings 8 , 9 favoring the use of movement sensors that collect far less identifiable data. Such movement sensors would typically collect tri-axial measures of both linear acceleration (accelerometer) and angular velocity (gyroscope) 10 , 11 , and they can be attached to infants individually 12 , 13 or by using more comprehensive multi-sensor suits 5 , 10 , 14 . Analysis of the data is traditionally based on straightforward quantification of the amount of movements (i.e., an actigraphy-type measure) 15 , while machine learning-based analysis enables more sophisticated and higher level interpretations of the movement data 10 , 16 .…”

“…Detection of infants’ carrying/holding is often attempted by applying movement thresholds 15 , 16 assuming that externally generated movements are larger than infant-generated movements. An alternative strategy is to measure co-incident activities in sensors attached to both the infant and the caregiver 13 , relying on the assumption that carrying/holding is reflected in the movements of both the infant and the parent synchronously. While the studies may report relatively high classifier performances with the given training datasets, there are as of yet insufficient means to measure holding/carrying behavior in unsupervised settings, such as out-of-hospital/lab without direct guidance by the research personnel.…”